English Translation of a Draft Summary on Fukushima's Thyroid Cancer Found in the Second-Round Screening

⁂ This is an unofficial English translation of the draft version of the official summary on the second-round results.
⁂ Update on November 10, 2019: The final version submitted to the 35th meeting of the Oversight Committee on July 8, 2019 had an addition of "and prospective" in the second to the last sentence as shown below.

   Further, in the future, estimated thyroid exposure doses should be used in case control and prospective studies with confounding factors adjusted, in order to evaluate an association between doses and thyroid cancer incidence rates.

＊＊＊＊＊＊＊＊＊     
     On June 3, 2019, the 13th Thyroid Examination Evaluation Subcommittee (herein, the TUE Subcommittee) was held, marking the end of a two-year term for the members. As it is customary for committees to produce a report at the end of the term, a draft version of the "interim report" was submitted by Gen Suzuki, chair to the TUE Subcommittee, concluding, "at this time, no association is seen between thyroid cancer detected in the second round and radiation exposure."
     The conclusion, albeit provisional, came as a surprise to many, because it did not seem like the TUE Subcommittee had spent enough time analyzing the second-round data. On the contrary, most of discussions have revolved around the topic of benefits and harm of the TUE itself.
     When this "provisional" conclusion became known to English speakers via a online newspaper article on the Mainichi website, confusion ensued. English translation of the "interim report" is offered below so everyone can see the "basis" for the conclusion by the TUE Subcommittee. Translation of some of the analyses mentioned is to be posted separately.

◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇◇
Note: This is a draft version. The final version is expected to have slightly different wording. (See the update at top.)

   A summary of the subcommittee regarding the results of the Full-Scale Screening of the Thyroid Ultrasound Examination (the second round)

June 3, 2019

Thyroid Ultrasound Examination Evaluation Subcommittee by the

Oversight Committee for Fukushima Health Management Survey

     The Thyroid Ultrasound Examination (herein, TUE) Evaluation Subcommittee by the Oversight Committee for Fukushima Health Management Survey (herein, the Subcommittee) released an “Interim Summary on the TUE” (herein, the Interim Summary) in March 2015 to report on the results of the Initial Screening of the TUE (the first round) conducted from FY 2011 to FY 2013. The Interim Summary concluded that the results of the first round “were unlikely to be the effects of radiation.”

     At the 26^th Oversight Committee for Fukushima Health Management Survey (herein, the Oversight Committee) held on February 20, 2017, a proposal was made to convene the Subcommittee in order to summarize and evaluate the results of the Full-Scale Screening of the TUE (the second round, FY 2014-2015). In response to this proposal, the 7^th Subcommittee was convened in conjunction with the 27^th Oversight Committee on June 5, 2017, followed by the 8^th Subcommittee held with a new set of members on November 30, 2017. The Subcommittee engaged in repeated deliberations over a total of 7 sessions, up to the 13^th Subcommittee held on June 3, 2019.

     Based on the contents of the deliberations, the Subcommittee summarized below the results of the second round, opinions of the Subcommittee on the results, and issues for future consideration.

1    Results of the Full-Scale Screening of the TUE (the second round)

     The second round that began in April 2014 covered approximately 380,000 Fukushima residents, adding subjects born between April 2, 2011 and April 1, 2012 to the first-round subjects born between April 2, 1992 and April 1, 2011. About 270,000 have participated as of June 30, 2017 (participation rates of 71% overall, 86.4% in ages ≤ 17, 25.7% in ages ≥ 18), and 2227 (0.8%) were assessed the B test results requiring a confirmatory examination and no one had the C test result (note: requiring an immediate confirmatory examination). After undergoing fine-needle aspiration cytology (FNAC) during the confirmatory examination, 71 were diagnosed with confirmed or suspected malignancy (26.2/100,000; 32 males and 39 females; average age 16.9 ± 3.2 years, age range 9-23 years; average age at the time of the accident 12.6 ± 3.2 years, age range 5-18; average tumor diameter 11.1 ± 5.6 mm). (Note: 52 underwent surgery, revealing 51 papillary thyroid cancer and 1 other cancer.)

     The detection rate of thyroid cancer during the first round was several orders tens of times multiples of ten higher than prevalence rate/proportion estimated from the Japanese cancer statistics data based on the regional cancer registry. The thyroid cancer detection rate in the second round was somewhat smaller than that in the first round but still several orders multiples of ten higher (than the cancer statistics data).

     Detection rates of confirmed or suspected cancer showed no regional difference in the first round. In the second round, a simple comparison without adjusting for sex and age showed the highest detection rate in the 13 evacuated municipalities, followed by Nakadori, Hamadori, and Aizu.

     However, detection rates of confirmed or suspected cancer are likely influenced by many factors. When some of the factors were considered, the following tendencies were observed.

• The detection rate of nodules with diameter of 5.1–10 mm was low in the 13 evacuated municipalities in the first round. Also, among subjects with the B test results in the second round, the proportion of the subjects whose first-round results were also B was low in the 13 evacuated municipalities. This suggests that the second-round results are affected by the first-round results.
• The longer the interval between the first and second rounds, the higher the FNAC rate and the detection rate of confirmed or suspected cancer. The longest average interval was observed in the 13 evacuated municipalities.
• FNAC rates have declined in successive years even in the first round. The second-round FNAC rate declined in the order of the 13 evacuated municipalities, Nakadori, Hamadori, and Aizu.
• When FNAC was conducted in the first round, FNAC rates and detection rates of confirmed or suspected cancer tended to be lower in comparison with the subjects who underwent no FNAC in the first round. 

2   Regarding the preliminary analysis on an association between thyroid cancer detection rates in the second round and radiation doses

     The above analysis has shown that detection rates of confirmed or suspected cancer are influenced by many factors other than sex and age at screening, such as screening year, FNAC rates, screening interval since the first round, and whether FNAC was conducted in the first round. Therefore, an evaluation of an association between thyroid cancer detection rates and radiation doses calls for analyses that control these factors.

     For radiation doses, an analysis was tentatively conducted with absorbed doses to the thyroid estimated in different age groups and municipalities by United Nations Scientific Committee on Effects of Atomic Radiation (UNSCEAR), which incorporate internal exposure.

     The analysis showed no clear association between doses and thyroid cancer detection rates.

          

3   Findings

     Proportions of the B test results in the primary examination, recommended to advance to the confirmatory examination, were larger with an older age at the time of the accident, and detection rates of confirmed or suspected cancer were higher with an older age at the time of the confirmatory examination. This is dissimilar to detection of many more thyroid cancer in a younger age group after the Chernobyl accident. Detection of more cancer with an increasing age is similar to cancer in general.

     The male-to-female ratio is close to 1:1, which is different from what tends to be observed in a clinical setting (about 1:6). It has been reported that the male-to-female ratio tends to be smaller in latent cancer and younger ages. Evaluation of an association between the male-to-female ratio and radiation remains as a future task.

     A simple comparison of detection rates of confirmed or suspected cancer in four regions appears to show differences, but it is influenced by many factors such as screening year and screening interval since the first round. Any analysis needs to take these factors into consideration.

     When factors that can influence detection rates were adjusted as much as possible, analyses of an association between doses and thyroid cancer detection rates, tentatively using the absorbed doses to the thyroid estimated by UNSCEAR by age group and municipality, revealed no consistent relationship between increasing doses and increasing detection rates (dose-effect relationship).

     Thus, at this time, no association is seen between thyroid cancer detected in the second round and radiation exposure.

4   Explanation to residents covered by the TUE

     The Subcommittee has advanced discussion on contents of the explanation to residents covered by the TUE. It is important to continue careful explanation of merits and demerits of the TUE, so their understanding and consent may be gained before the TUE is administered.

5   Viewpoints for future evaluation

     The third round started in FY 2016 and the fourth round in FY 2018, and their results need to be incorporated into further analyses.

     Also, participation rates of the TUE are declining every year. In particular, a participation rate after high school graduation is low with continuing decline expected in the future. Moreover, it is possible that more cancer cases might be discovered outside the TUE conducted as part of the Fukushima Health Management Survey.

     This leads to the use of regional and national cancer registries in order to identify cancer incidence among residents covered by the TUE so an analysis could be conducted.

     Further, in the future, estimated thyroid exposure doses should be used in case control studies with confounding factors adjusted, in order to evaluate an association between doses and thyroid cancer incidence rates.

     These are the viewpoints that should be considered in advancing discussion at the Subcommittee and the Oversight Committee in the future.

      

A draft summary of the subcommittee regarding the results of the Full-Scale Screening of the Thyroid Ultrasound Examination (the second round) by on Scribd

Pediatric and Adolescent Thyroid Cancer Data: Fukushima and Beyond

*Table for Latest Data was replaced with a corrected version showing that Age 25 Milestone Examination was conducted in FY2017, not FY2016 as stated in the original version. Also a link was added for the post on the unreported cases. (September 28, 2019)

Thyroid Ultrasound Examination (TUE)
     Eight years have passed since the March 11, 2011 triple disaster of earthquake, tsunami, and nuclear accident in Japan. The Thyroid Ultrasound Examination (TUE) was launched in October 2011 as part of the Fukushima Health Management Survey (FHMS), targeting over 367,000 children who were aged ≤ 18 and residing in Fukushima Prefecture at the time of the accident. The FHMS was commissioned to Fukushima Medical University (FMU) by Fukushima Prefectural government, with the fund originating in Ministry of Economy, Trade and Industry (METI) which manages nuclear power plants. The fund has been distributed by Ministry of the Environment (MOE) which oversees environmental contamination by toxins such as heavy metal, asbestos, and radiation. Incidentally, involvement of Ministry of Health, Labour and Welfare (MHLW) with radiation exposure issues is limited to nuclear workers (worker's compensation certification and epidemiological studies) and atomic bomb survivors (medical certification).
     Details of the TUE are described in the September 2017 Fact Sheet. The first-round screening, which was scheduled to run from October 2011 through March 2014, was considered to provide "baseline" prevalence of thyroid cancer in Fukushima children aged 18 or younger during the "latency period." This was based on the fact that thyroid cancer dramatically increased in young children beginning 4 years after the 1986 Chernobyl accident. The first round was to be conducted through Fiscal Year (FY) 2013, ending in March 2014 with a Japanese FY running from April through the following March, and each subsequent round every two years. 

Time lag and overlapping data
     In reality the primary examination (cursory ultrasound screening) of the first- round screening dragged on for an additional 13 months, overlapping with the first year of the second-round screening. This extended period was likely due to a simple lack of manpower and a large number (> 300,000) of participants. Furthermore, if someone decided to undergo the TUE for the first time during the time the second round screening was going on, they would be included in the first-round cohort as long as they had not received notifications for the second-round screening. (Part of the reason was probably to increase a participation rate, although FMU officials insist they were honoring the "right of residents to participate in the first-round screening.") 
     With the primary examination lagging behind, the confirmatory examination for those who had ultrasound abnormalities also lagged behind, extending way beyond each screening period, with an average wait time of 6 months. After undergoing a more detailed ultrasound examination and blood/urine testing, most confirmatory examination participants were placed on regular, biennial screening cycle, while some who needed more frequent observation were moved to a clinical follow-up track which was no longer part of the TUE as described in the next section. And yet some underwent fine-needle aspiration cytology (FNAC) which eventually detected 116 nodular lesions suspected of thyroid cancer in the first round: an unexpectedly high number. 
     In April 2018 the TUE began its 4th cycle of screening. The TUE data as of December 2018 show that cytologically suspected thyroid cancer was detected in 212 patients, and 169 of them have undergone surgery which confirmed thyroid cancer in 168 and benign nodule in 1. 

Missing data
     We now know that this is far from a complete picture, as the official data released by FMU to the quarterly FHMS Oversight Committee only includes thyroid cancer cases diagnosed directly during the confirmatory examination. As mentioned above, some confirmatory examination participants are placed on a clinical follow-up track for closer observation. Unlike the specially-funded TUE, the follow-up visits are considered regular medical visits, which are covered by the national health insurance. A more private nature of the follow-up visits gives FMU a cover not to release clinical data: the follow-up track has effectively become a "black box." On a rare occasion, a case "pops out" of the black box, as in thyroid cancer diagnosed in a male who was at 4 at the time of the accident (see this post). 
     Moreover, an official report has revealed that, as of June 2017, there are 12 thyroid cancer surgeries conducted at the FMU hospital which have never been reported to the Oversight Committee, including 3 cases that never participated in the TUE. (See this post for English translation of the report.) The table shown below reflects these 12 "unreported" cases. Also, FMU has admitted not to have kept track of surgeries conducted at facilities other than the FMU hospital beyond the already known 7 cases from the first round, making the actual number of surgically confirmed cases uncertain. 


Latest Data

(See this post for details of the "unreported" cases.)

The TUE support program as a source of supplementary information
    The TUE support program, as described in this post, is offered by Fukushima Prefecture to offset the out-of-pocket expenses (30% co-payment) associated with receiving medical care for thyroid nodular lesions suspected to be malignant, in exchange for clinical information. Eligibility criteria limit benefits to those not receiving any other medical or public assistance, which means those aged ≥18 because those under age 18 receive free medical care in Fukushima Prefecture. At age 18, many residents leave Fukushima Prefecture for schooling and jobs, and the TUE has a very low participation rate for this age group. In addition to clinical information, the TUE support program could supplement the official data with basic data of participants undergoing the TUE and subsequent medical care elsewhere. However, very little effort has been made to analyze the collected information to compare with the FMU data.
     For instance, according to the most recent report, the support program disbursed 375 payments to 257 individuals through December 2018, including 95 surgeries for 93 recipients. Pathological diagnoses for these 93 individuals include 87 thyroid cancers (85 papillary, 1 poorly-differentiated, and 1 follicular) and 6 non-cancer cases such as follicular adenoma. Knowing how many of these surgical cases are included in the FMU data would help gain a better grasp of the actual thyroid cancer status.

Non-transparent data
     Besides the "missing data," Fukushima's thyroid cancer data suffers from lack of transparency, hindering third-party analysis and even discussions/interpretations at the Oversight Committee and the Thyroid Examination Assessment Subcommittee (herein, the TUE subcommittee).
     First, very little clinical information was initially released, confusing and complicating  discussions on the high number of cancer cases detected and whether unneeded surgeries were being conducted. When the clinicopathological data on surgical cases was first released in November 2014 (see this post for translation), it was neither specific nor adequate enough to create a more integrated clinical picture for each case. The format remained essentially the same even in the "latest" update in September 2016.
     Second, the format in which data is released has actually "devolved." From the third-round screening onward, neither the results nor the implementation status of the confirmatory examination are no longer available by municipality. Instead they are only offered by four "administrative" regions. FMU explains this shift was to avoid potential identification of individual cancer cases particularly in small municipalities, but the shift to provide less detailed data would effectively "shut down" a continued analysis by Tsuda et al from Okayama University.

Sharing data
     FMU's Radiation Medical Science Center website declares "unshakable resolve to disseminate survey results domestically and internationally."  This resolve mostly appears to concern the world of academia, such as journal publications and annual meetings of thyroid and endocrine organizations in Japan and overseas. Time after time, academic presentations have taken precedence over presenting data and its analysis to the Oversight Committee and/or the TUE subcommittee, let alone residents. 
     For instance, the September 2016 clinicopathological data (with data compiled as of March 2016) mentioned above was first released in English at an international symposium. (Note: This blog posted some of the data presented in English and Japanese in October 2016.) This data was not officially released in Japanese until November 2017, when the TUE subcommittee resumed after a three-year hiatus with new members. However, by this time, a more updated version of data compiled as of March 2017 had been presented at the Japan Thyroid Association meeting in October 2017. This data was also presented at the American Thyroid Association meeting in October 2018, as described in this post under "Reluctance on releasing detailed data," but it has yet to be presented to the TUE subcommittee.
     Publications in English are beyond the reach of most residents whose very data are presented in them, due to a language barrier and frequent paywalls. Although most are eventually summarized in plain language on the Japanese website of the Radiation Medical Science Center for FMU, there is a time lag and definite information gap.
     Japanese summaries of selected papers have been presented to the Oversight Committee or TUE subcommittee meetings. These tend to be in more detail. and papers selected tend to have "favorable" conclusions supporting "absence" of radiation exposure.
   
Second round analysis
     With the first-round screening designated as "baseline screening," data from the second-round screening onward were to be carefully analyzed to assess any effect of radiation exposure. Yet, even now, no "proper" analysis of the second-round data has been conducted when the fourth-round screening is ongoing and the third-round results are about to be finalized. This is mainly because the TUE subcommittee, which was supposed to be conducting such analyses, took a three-year hiatus for unknown reasons. 
     When the TUE subcommittee resumed with new members in November 2017, FMU began to submit various "analyses" of the second-round data but they have had limited value because they were uninformative due to insufficient information (i.e., data shown in percentages rather than actual numbers) or unreliable due to unclear methodology (as analyzed by Junichiro Makino in April to June 2019 issues of Kagaku, only available in Japanese).

Cancer screening vs. health monitoring
     The English website of FMU's Radiation Health Science Center states that the primary purpose of the FHMS is to "monitor the long-term health of residents, promote their well-being, and confirm whether long-term, low-dose radiation exposure has health effects." 
     Hence the TUE is not considered "cancer screening" in a traditional sense, which measures its effectiveness by reduced mortality. (Thyroid cancer screening in adults is not recommended because screening does not lead to reduced mortality.) 
     Rather the TUE is considered "health screening" to monitor health and well-being of residents against a unique backdrop of radiation exposure. However, the TUE is often considered in traditional context of cancer screening, leading to misguided and confusing discussions as described in the next section.

Overdiagnosis? 
     The official discourse of the TUE leans towards "overdiagnosis" of thyroid cancer, which implies that these cancers are "quiet" cancers which will never have been clinically found for lifetime without the TUE. However, this discourse is incongruous with clinical data which shows 39.2% extrathyroidal spread, 77.6% regional (lymph node) and 2.4% distant (lung) metastasis. As "health screening," the TUE is leading to early diagnosis and treatment of thyroid cancer, requiring less aggressive surgery and treatment and maintaining decent QOL.
     Yet some Oversight Committee and TUE subcommittee members insist that detection of these cancer cases could have waited until becoming symptomatic instead of being detected by screening, because the existing literature show that the survival rate of thyroid cancer in children is very high even with lymph node metastasis. However this claim should be examined cautiously. 
     Pediatric thyroid cancer is rare, and most of the existing studies on survival rates are from outside Japan. Clinical management strategies of pediatric and adolescent thyroid cancer in Japan do not necessarily follow those recommended in the 2015 ATA pediatric guidelines. In Japan, a more conservative surgical method, hemithyroidectomy, has been used partially because of limited availability of radioiodine treatment. Also in Japan lymph node dissection is conducted more frequently and extensively. 
     On the other hand, clinical management strategies in other countries gear towards total thyroidectomy with limited lymph node dissection, often followed by radioiodine treatment that kills any remaining cancer cells. This might lead to a high survival rate, but it is not without any side effect and/or complications which might lower QOL.
     It is impossible to know what would have happened to the Fukushima cases without surgery. For record, surgery was indicated for the vast majority of them. It is certainly inappropriate to extrapolate findings from different treatment approaches in order to negate the necessity of the TUE. 

Difficult comparison of data
     Interpretation of the Fukushima data is difficult due to the absence of comparative data in Japan. The MOE sponsored three-prefecture study conducted in supposedly non-exposed Aomori, Yamanashi, and Nagasaki Prefectures does not offer a valid comparison due to non-matched cohort and lack of statistical power. 
     A straight comparison between Fukushima data and other existing data is even more difficult. One of the reasons is lack of other large-scale screening data in this age group in populations with a similar iodine status. Another reason, previously described, is a difference in surgical methods and treatment approaches which complicates comparison of metastatic rates.
     Regarding extrathyroidal extension or invasion, Japan has its own classification, "Ex." So far all Fukushima cases have been classified as "Ex1" meaning minimal extension and corresponding to "T3" in the TNM classification (in the previous 7th edition of AJCC/TNM staging system). Other data such as a recent study from Czech Republic do not always specify degree of extension, not keeping T3 and more extensive T4 (= "Ex2") separate. 

Addendum on June 7, 2019:
     The 1998 study by Nagasaki University, "Childhood thyroid cancer: comparison of Japan and Belarus," including Yamashita and late Nagataki as co-authors, offers some data in pediatric thyroid cancer which were diagnosed clinically in Japan. 
     As seen below, in the Japanese clinical cases, mean age at operation was 11.9 ± 1.9 years (17.8 ± 3.1 years in Fukushima) and mean tumor size was 41 ± 17 mm (14.0 ± 8.5 mm in Fukushima). Regional lymph node metastasis was found in 73% (77.6% in Fukushima), pulmonary metastasis in 19% (2.4% in Fukushima), and extrathyroidal extension in 35% (39.2% in Fukushima). (Fukushima data can be seen here.)

Updated staging system
     Complicating the picture even more, the 2017 update to the 8th edition of the AJCC/TNM staging system no longer includes lymph node metastasis in T3 or considers minimal, microscopic extension in staging, potentially leading to downstaging. 
     Thyroid cancer in age < 55 never exceeds stage 2 by the current staging system. However, for children and adolescents, there is no single staging system that is appropriate as stated in the aforementioned ATA pediatric guidelines. Furthermore, a disparity in classifying extrathyroidal extension between the Japanese Ex system and the 8th edition of AJCC/TNM could lead to difficulty in discerning the Fukushima data.

Valuable but deceitful data
     Because ultrasound screening for thyroid cancer is not commonly conducted in children and adolescents, the TUE data from Fukushima has been welcomed as a valuable source of epidemiological data for lesions such as thyroid nodules (benign and malignant) and cysts as well as ectopic intrathyroidal thymus.
     One cannot fault those embracing the Fukushima data for not being aware of the fact that the data is ladened with incompleteness and lack of transparency. After all, the data comes from FMU, supposedly an authoritative source, so what is there to doubt about? 
     It takes more than careful discernment, namely Japanese language skills, to identify inconsistencies or inadequacies in the information released by FMU, even in peer-reviewed journals. In fact, most of the studies published by FMU, including some published collaboratively with other institutions such as Nagasaki University, tend to be biased, as explained later.
     Publishing studies with incomplete dataset should be scientifically unacceptable. But that is what's happening with the TUE papers, and there has never been official disclosure of missing data in English. 
     
Inconsistent data
     It's not just that data is missing but sometimes data presented is inconsistent. One such example is a 2018 paper published in Thyroid by Yamashita et al, which is the most recent "comprehensive" TUE paper at this time. It is stated in this paper, "Out of 146 surgical cases, 126 patients underwent surgery at Fukushima Medical University Hospital, and 125 were postoperatively diagnosed with thyroid cancer," which suggests there are 20 patients that underwent surgery at facilities other than FMU. That's not the case.
     The number of surgical cases, 146, is from data as of December 2016, while the number of patients undergoing surgery at FMU, 126, is from data as of March 2016 when there were 132 surgical cases. Of 126, one was postoperatively diagnosed as benign nodule, leaving 125 with thyroid cancer. The known number of patients undergoing surgery at facilities other than FMU is and remains 7, as mentioned in "Missing data" section, which means 139 of 146 surgical cases would have been diagnosed with thyroid cancer after undergoing surgery at FMU. 
     Also Table 2 shows the total number of cases suspected of thyroid cancer for the first through third rounds, 191, which is from data as of March 2017 when the number of surgical cases was 153 (including one benign case). All these numbers are so confusing that most readers will not have any idea that the data presented in the most recent TUE paper is inconsistent and partially outdated. And certainly the way the surgical data is presented does not suggest how outdated it might be.
    The Yamashita paper, like other FMU papers by Suzuki (2016) and Suzuki et al (2016), gives an impression that the "baseline" actually covers the first 4 years after the accident, not the first 3 years when the first-round screening was conducted, obscuring the line between the first and second rounds. In reality, the line did become fuzzy during the fourth post-accident year when the first and second rounds went on simultaneously with some new cases still being added to the first round, but that should not lead to an altered baseline. 
     This arbitrary shift of baseline is accompanied by dismissal of radiation effects based on comparison with Chernobyl (no cancer case in age ≤ 5, lower radiation doses) and claim of no regional difference (i.e., dose-response). The latter, absence of regional differences, has solely been based on analyses of the first-round data only by Ohira et al (2016, 2018), with questionable classification of municipalities into dose groups not necessarily reflecting actual doses. 
     In addition, due to insufficient internal dose data, the analyses were conducted with external doses which are not as relevant. These external dose estimates are derived from the Basic Survey portion of the FHMS, with only a quarter of the residents returning the individual questionnaire used in estimating doses. Dose estimation by nature is accompanied by large uncertainties, which are further exacerbated by recall bias inherent in a questionnaire-based survey.
     Thus the TUE papers by FMU have consistently clouded the picture by obscuring the baseline boundary and prematurely dismissing radiation effects even before the second-round data is properly analyzed. 
     
Misrepresented data 
     Then, FMU went one step further by shifting the baseline to 5 years. 

     On November 29, 2018, another FMU paper was published by Ohtsuru et al in JAMA Otolaryngology-Head & Neck Surgery, evaluating the number of thyroid cancer cases by age groups within 5 years of the Fukushima accident. Comparing basic clinical characteristics and demographic patterns in the first and second rounds, the Ohtsuru study claims similarities between the two rounds with no major changes in overall characteristics within 5 years, suggesting that the first 2 rounds simply uncovered a "subclinical pool of thyroid cancer" in children and adolescents and implying that the first 5 years after the accident constitute a de facto baseline.
     Unbelievably, even before the second-round data is properly analyzed to examine any potential effects of radiation exposure, thyroid cancer cases detected during the second round are clumped together with cases from the first round to constitute a "subclinical pool of thyroid cancer," previously unknown in children and adolescents.
     Although it is likely for some of them to be subclinical cancers whose growth might have been promoted by radiation exposure, there really is no way to know which ones. Of 125 surgical cases, 5 turned out to be microcarcinoma (< 10 mm) with no extrathyroidal extension or metastasis (lymph node or distant). Lack of more relevant clinical information does not allow any further scrutiny. This is also a delicate subject: calling only some of the detected cancers subclinical is divisive, giving rise to a sense of inequality especially involving any future potential compensation.
     It is unscientific to collectively designate all thyroid cancers detected during the first and second rounds as "subclinical cancer." But it happens to nicely explain away why 58 of 71 thyroid cancer cases detected in the second round had no ultrasound findings in the first round: subclinical cancers only became "detectable" with time.
     Ohtsuru et al. back up this claim of a "subclinical pool of thyroid cancer" with essentially 3 things: autopsy data and two FMU studies by Midorikawa et al (2017) and Takahashi et al (2017). As these 3 elements play an important role to support the claim of "overdiagnosis of subclinical cancer" in Fukushima, each is reviewed in detail below.

1) Autopsy data
     Ohtsuru et al state, "many more thyroid papillary cancers are observed in autopsies among people who are relatively young than among those in cancer registries," citing two papers, a 2016 meta-analysis and a 1984 Swedish study. What is meant by "relatively young" is vague here. Fukushima's thyroid cancer cases have been detected in ages up to 25 so far. 
     In the meta-analysis, there is one study with median age of 16 with 13 thyroid cancer cases. However, in this 1986 Finland study by Franssila and Harach based on autopsies conducted on 93 cases (age < 40), only 3 of 13 thyroid cancers were detected in ages 11–20 (n = 22) with the youngest being age 18. No cancer was found in age < 1 (n = 15) or ages 1–10 (n = 21), while 4 cases with unspecified age were detected in ages 21–30 (n = 18). 
     The 1984 Sweden study by Bondeson and Ljungberg, also included in the meta-analysis, conducted autopsy on 430 patients (ages 11–100) showing 2 cancer cases (both age 17) in ages 11–20 (n = 12) and 1 case (age 21) in ages 21–30 (n = 50). 
     The Finland study goes so far as to say, "Although OPC (occult papillary cancer) seems to be rare in children, its prevalence in young adults after age 18 years is the same order as in older age groups, although there may be a slight rise in prevalence at middle age." 
     In Fukushima, about half of those with nodules suspicious of cancer are age <18 (at diagnosis) overall, but the proportion of age < 18 increases with subsequent rounds: 43% in the first round, 55% in the second round, and 57% in the third round. Although autopsy data might suggest presence of subclinical thyroid cancer in the age ≥ 18 (or age ≥ 17 considering the Swedish study) group, it certainly could not designate all the Fukushima cases in that age group or other age groups as "subclinical" without consideration of clinicopathological characteristics of individual cases.

Addendum on June 7, 2019:
     The 1986 Finland study is cited in "Histopathological characteristics of pediatric thyroid cancer in Gomel, Belarus" by Ito et al (1996) as evidence that "occult thyroid carcinoma is observed rarely at autopsy in adolescents." This is the opposite of how the Finland is study is cited by FMU, an interesting fact especially because co-authors of the 1996 study include Yamashita and late Nagataki, key players after the Fukushima accident. 

2) The Midorikawa study
     In the TUE, all participate in the primary examination with cursory ultrasound screening, and only those with nodules ≥ 5.1 mm or cysts ≥ 20.1 mm undergo the confirmatory examination with a more detailed ultrasound examination as well as urine and blood tests. Midorikawa et al compared two ultrasound measurements from the primary and confirmatory examinations in 116 thyroid nodules suspected of cancer in the first round, classifying them into 3 groups according to changes in tumor diameter (less than - 10%, ± 10%, greater than + 10%). 
     As a result, 7 cases with an average diameter of 12.1 ± 5.2 mm showed an average of 16.5 ± 4.8% reduction in size, 81 with an average diameter of 13.9 ± 8.5 mm showed an average of 1.7 ± 5.5% increase in size, and remaining 28 with an average diameter of 12.8 ± 7.9 mm showed an average of 24.1 ± 24.9% increase in size.
     Midorikawa et al also used mathematical models to evaluate the tumor growth rate, concluding that a tumor with a diameter of 5 mm would reach a growth arrest point in 8.0 (range 5.1-17.6) years. 
     In Discussion, the authors state that many of thyroid cancers in the first round "did not grow linearly and instead entered growth arrest, even in young patients. Furthermore, the mean duration during which the tumor grew from a diameter of 5 mm to the growth arrest point was 8.0 years."
     There are two issues with this statement. First, the average observation period of barely 6 months is rather short to declare the cancers "entered growth arrest." Second, the statement is written as if the growth arrest point was actually "observed." This is misleading because the growth arrest point is based on simulation.

     There are other issues in this study, such as questionable methodology. 

     A criterion of "10%" was selected because Choi et al suggest, "Any differences smaller than 13.1% and 7.3% in volume and maximum diameter, respectively, measured by using US for well-defined thyroid nodules of > 1 cm should not be considered as a real change in size." Ultrasound diagnostic criteria for malignant nodules include irregular shape and ill-defined edge with jagged border, and Ohtsuru et al have shown in Table 1 that 55 of 116 cancer cases In the first round had tumor diameter of 5.0–1.0 mm. 
     As addressed by the authors, changes in a tumor diameter do not accurately reflect changes in tumor volume or tumor cell proliferation. It is curious why the authors did not use the actual tumor volume and heed the caution from Choi et al not to rely on volume change of smaller than 13.1%, especially when half the authors were involved in a 2015 study that actually determined thyroid volume based on the elliptical shape volume formula.   
     Further, unlike the authors' suggestion, changes in tumor diameter alone are clearly not adequate in determining a treatment course, considering most of these cases actually had valid indications for surgery. Linking the observed changes with clinical behaviors of each tumor, when available, would have been much more informative. 
     Last but not the least, this study offers an excuse not to look for tumor that is going to stop growing anyway, a good news to Midorikawa who is known to actively discourage participation in the TUE in order to reduce a psychological burden. 
    
3) The Takahashi study
      In the Takahashi study, a cancer progression model was formulated under the assumption of no radiation exposure, and the number of pediatric and adolescent thyroid cancer cases that would have been detected in Fukushima was estimated based on the national cancer registry (NCR) data with a sensitivity simulation.  
     In other words, the study estimated how many thyroid cancer cases would have been diagnosed by the TUE in Fukushima if radiation exposure had never occurred. It concluded that suspected thyroid cancer cases detected in 77 females and 39 males during the first round fell within 95% confidence interval of the estimates derived from the model. (Translation: There was a 95% chance that the same number of cancer cases would have been detected from the TUE without the accident, a finding that essentially dismisses radiation effects on the first round results.)

     The Japanese summary presented to the 8th TUE subcommittee (held on January 26, 2018) states, "(...) parameters for the model were estimated so that the estimated prevalence matches values from the national cancer registry. As a result, latency was estimated to be 34 years for male and 30 years for female (...)." This statement suggests that Takahashi et al found just the right parameter values so that the estimated prevalence would match the first round results 95% of the time. 
     The latency (called "sojourn time" in the study) of 34 years for males and 30 years for females means that the cancers would not have become clinically apparent for 30-34 years without the TUE. The authors appear to equate latency with age, assuming latent cancer already exists in infancy, which is a scientifically unfounded assumption claimed by some researchers, most notably Toru Takano of Osaka University. (Note: Takano is a controversial figure who happens to be the only dual member of the Committee and TUE subcommittee upon recommendations by the Japan Thyroid Association.)  

     A sensitivity simulation was conducted in order to overcome 3 difficulties encountered when directly comparing the TUE data with the NCR data: 1) different index (prevalence proportion in the TUE vs. incidence rate in the NCR), 2) method of detection (mass screening in age 0-18 in the TUE vs. routine detection in clinical settings in all ages in the NCR), and 3) unknown sensitivity in the TUE. 
     However, accuracy of the NCR data itself is in question with a poor reporting quality with the overall DCO (Death Certificate Only) rate of 13.4% (should be < 10% according to the international standard). Also, cancer registry in Japan only became mandated in 2016, and the 2009 NCR data used by Takahashi et al. is much lower in quality than the TUE data. In addition, detection method was unknown in a little over 80% of thyroid cancer in the NCR.
     Poor accuracy and unknown sensitivity of the NCR data question validity of the authors' statement, “In this research, we only used published NCR data, and we built a radiation-free model, which is applicable in any region in Japan.” 

Endorsing the Ohtsuru study
     The Ohtsuru study has essentially shifted the baseline period from 3 to 5 years, based on a claim that the cases detected in the first and second rounds constitute a "subclinical pool of thyroid cancer." A careful review of the "supporting evidence" proves that this is an unfounded claim. Yet the claim gains seemingly legitimate support in Invited Commentary accompanying the Ohtsuru study, co-authored by Bauer and Davies who were members of TM-NUC, an IARC international expert group on thyroid monitoring after nuclear accidents. (See this post for more details on TM-NUC.)
     Not only do Bauer and Davies erroneously acknowledge that the TUE data from the first 5 years after the accident represents "background rates of cancer," but they also praise Ohtsuru et al for "doing an excellent job" explaining why those Fukushima cases are "prevalent (subclinical, asymptomatic) disease." Proceeding to highlight two points addressed by Ohtsuru et al, latency and molecular signature, Bauer and Davies do an excellent job of self-revealing their lack of knowledge in radiation-induced thyroid cancer in children, which is alarming considering they are the expert members on TM-NUC. 

     After acknowledging that latency for radiation-induced pediatric thyroid cancer was 3-5 years after Chernobyl, Bauer and Davies propose that latency is longer among iodine-sufficient people at 5 to 10 years, citing Annex A of the UNSCEAR 2016 report. They conclude that Japan is an iodine-rich country and hence latency for Japanese people is longer at 5 to 10 years, "which is longer than the 2-year and 4-year post-accident time frame in which these baseline measures were obtained." (Notice they get the time frame wrong here: It should be "the 3-year and 5-year post-accident time frame.") This effectively gives the scientific-sounding backing to the arbitrary shift of baseline in the Ohtsuru study.
     There are several issues with this claim. First, even though a traditional Japanese diet includes high-iodine foods, not all Japanese people at present time are iodine-sufficient owing to modern diet and life style, as shown in a urinary iodine study and acknowledged in comments by other TM-NUC members. Second, there is no evidence for longer latency for iodine-sufficient people in UNSCEAR 2016 Annex A. An email inquiry to Davies, corresponding author, revealed that evidence would be found in the Life Span Study (LSS) of the atomic bomb survivors. However, nothing of the sort was found in the LSS papers. 
     The truth is that the LSS does not offer a good basis for latency for thyroid cancer because the tumor registry was not established until 1958. (To be exact, 1957 in Hiroshima and 1958 in Nagasaki.) In other words, there isn't good data on any cancer in the first 12-13 years after the atomic bombs were dropped, and the LSS data does not offer evidence for latency of 5 to 10 years. Also, Atomic Bomb Casualty Commission (ABCC), now Radiation Effects Research Foundation (RERF) was initially alerted to leukemia, not thyroid cancer.

     Next, Bauer and Davies note a higher incidence of BRAF point mutation and fewer cases with RET/PTC gene rearrangement in Fukushima than in Chernobyl. As stated in a title of the cited study, this is a "different oncogenic profile from Chernobyl," seemingly validating FMU's claim that Fukushima cases are not radiation-induced.
     However, as Gerry Thomas writes in Chapter 12 of Thyroid Cancer and Nuclear Accidents, it is now known that "RET rearrangement and BRAF mutation are not related to exposure to radiation, but show a strong association with age of the patient at operation." Thus a "different oncogenic profile from Chernobyl" is nothing more than a reflection of a "different age profile in Fukushima."

     To be fair, Bauer is a pediatric endocrinologist and Davies is an ENT surgeon who may not necessarily be well-versed in the subject of radiation and thyroid cancer after nuclear accidents, at least not beyond what was learned as TM-NUC members. It does not help that FMU seems to do everything possible to obfuscate the situation by publishing studies after studies with arbitrary data and study design while withholding critical data from the public as well as third-party researchers.
     Even then, lack of discernment exhibited by Bauer and Davies is concerning. Most of all, their endorsement of the first 5 years as baseline, backing it up with an unfounded claim of "longer latency in iodine-sufficient people," is frankly misleading and alarming. 

     Then the plot thickens.

Beyond Fukushima
     On April 23, 2019, a study by Bernier et al was published in Cancer, showing that pediatric thyroid cancer in the US increased 4.43% annually from 1998 to 2013, and that this was not just because of increased surveillance but a "true increase." The Bernier study had actually been submitted on February 20, 2018 and accepted on May 8, 2018. When finally published a year later, it was accompanied by two editorial comments, one of which was co-authored by Chen and Davies. Aside from being an ENT surgeon, Davies is an overdiagnosis expert, who has done contract work related to overdiagnosis for the US Preventative Health Task Force (as per Davies's COI disclosure in the editorial comment) and co-authored multiple papers with Welch who is well known for his overdiagnosis theory. 
     The editorial comment was submitted on July 22, 2018 and accepted on December 19, 2018 while Davies was actively involved with TM-NUC. Apparently taking a full advantage of Davies's newly acquired knowledge about the Fukushima data, Chen and Davies refute Bernier et al for reporting that there is no reservoir of clinically silent cancers in the pediatric population. They state,

Because the radiation exposure was so low, the cancers were found so soon after the accident, and the growth patterns suggested the identified cancers actually were falling into an arrest pattern, it now is believed that the “high prevalence of childhood thyroid cancer detected in this four year study in Fukushima can be attributed to mass screening. The work from Fukushima strongly refutes the premise that there is not a reservoir of clinically silent cancers in the pediatric population. 

The reference to "the identified cancers actually were falling into an arrest pattern" obviously comes from the Midorikawa study which was available when the comment was written. This editorial comment was written before the Ohtsuru study was published, hence the reference to "this four year study in Fukushima," which is inaccurate but consistent with the cited studies as explained in section "Inconsistent data."
     Overall, Chen and Davies strongly point to overdiagnosis as a cause of the increase in pediatric thyroid cancer in the United States.

     What is curious is the publishing timeline of the Bernier paper. Why was it held up for nearly a year before actually getting published? Timeline of the accompanying editorial comments might give us a clue. The Bernier paper was received on February 20, 2018, revised on April 9, 2018, and accepted on May 8, 2018. The Chen and Davies comment was received on July 22, 2018 and accepted on December 19, 2018. The other comment by Goldenberg, an ENT surgeon, was received on November 1, 2018 and revised and accepted on March 19, 2019. Then all three were published on April 23, 2019.
     Countering the comment by Chen and Davies, the Goldenberg comment encourages investigation into "whether changes in environmental factors or lifestyle changes are driving part of this increase." Goldenberg further states, 

(...) it is our role as physicians to protect our patients from complacency and undertreatment. Explaining away thyroid cancers as being subclinical or clinically insignificant is reminiscent of days past when we told our patients: “don’t worry, it’s good cancer.”

   It appears that the Bernier paper was "held up" until two opposing comments were obtained. That wasn't the case with another study on pediatric thyroid cancer in the United States by Qian et al, which was accepted on March 24, 2019 and published in JAMA Otolaryngology-Head & Neck Surgery on May 23, 2019. 

     Like Bernier et al, Qian et al found a true increase in pediatric thyroid cancer. Both Qian et al and Invited Commentary by Rastatter et al acknowledge the Ohtsuru study. Qian et al state, 

Ohtsuru et al showed that large scale mass ultrasonography screening of children and young adults within 5 years of the 2011 Fukushima Daiichi nuclear power station accident identified many subclinical thyroid cancers in an age-dependent manner. (...) These high incidence rates represent prevalent disease rather than effects of radiation because all diagnoses were made before the effects of radiation could manifest according to epidemiologic evidence for postradiation-exposure thyroid cancer in Japan.

     Rastatter et al state, "(...) clinicians should be mindful that large-scale screening by ultrasound can identify nonclinical or subclinical thyroid cancers in children, putting them at risk for overdiagnosis." Unlike Chen and Davies, however, Rastatter et al do not attribute the increase to overdiagnosis. Rather, they encourage continued efforts to determine whether there is a true increase exists and what factors might be contributing to the increase.

TM-NUC
     There is no doubt that the TUE data (and actually other FHMS data) propagated by FMU are biased. Focusing on arbitrarily selected data and tactfully withholding relevant clinical data, FMU has successfully promoted an unfounded notion that all thyroid cancers detected in the TUE have existed subclinically prior to the Fukushima nuclear accident. This in turn offered new evidence of "a subclinical reservoir of pediatric thyroid cancer" to the world.

     The plot thickens even more.

     The IARC expert group TM-NUC recommends against systematic thyroid cancer screening after nuclear accidents for fear of overdiagnosis. (See their summary for details.) But evidence used in section "Overdiagnosis in Pediatric Thyroid Cancer" of the TM-NUC report 1 is the very evidence that was used in the Ohtsuru study as the backing for "a subclinical reservoir of pediatric thyroid cancer": the autopsy data as well as the Midorikawa and Takahashi studies. (Also mentioned is the Suzuki study which appears as Suzuki (2016) in section "Inconsistent data." Suzuki prematurely concludes that the Fukushima cases are unlikely to be due to radiation exposure, only one year into the second round when 25 of eventual 71 cancer cases were detected.) 
     Granted, evidence on the subject of overdiagnosis in pediatric thyroid cancer is scarce. But drawing "evidence" from the biased FMU studies only worsens the situation in two ways: 1) propagate the biased evidence and 2) legitimize the biased evidence. Incorporation of biased evidence into an official report of an authoritative international agency practically endorses such biased evidence. Whether the TM-NUC members were aware of the biased nature of the FMU studies is beyond anyone's guess. Looking at the member roster, some are usual suspects who might tend to overlook potential flaws and biases for their own gain. Others, like Davies, might be keen on grabbing some "evidence" from an authoritative source that would advance their own agenda.

     In late November, two months after the TM-NUC report 1 was released, JAMA Otolaryngology-Head & Neck Surgery published Clinical Guideline Synopsis on Thyroid Cancer Screening after Nuclear Accidents. Published in the same issue were the Ohtsuru study titled "Incidence of Thyroid Cancer Among Children and Young Adults in Fukushima, Japan, Screened With 2 Rounds of Ultrasonography Within 5 Years of the 2011 Fukushima Daiichi Nuclear Power Station Accident" and Invited Commentary by Bauer and Davies, titled "Why the Data From the Fukushima Health Management Survey After the Daiichi Nuclear Power Station Accident Are Important." (Incidentally, Davies is an associate editor of this journal.) These titles seem "loaded" once the dots are connected. 
     New "scientific evidence" based on the skewed Fukushima data was introduced to the world with an endorsement of "experts" who served on the IARC expert group TM-NUC. This new "evidence" of "a subclinical reservoir of pediatric and adolescent thyroid cancer" can attribute the recently reported increase in pediatric thyroid cancer in the United States to overdiagnosis. It is sobering to think this is the lesson from Fukushima.
     TM-NUC was fully sponsored by the Japanese government which shelled out 362,400 Euro (> $400,000), a bargain price to gain an endorsement of an authoritative international agency on "new baseline" which, for all practical purposes, dismisses potential radiation effects on Fukushima's thyroid cancer.
     
Internal review or censorship?
     Why do studies on the FHMS data published by FMU, either alone or collaboratively with other institutions such as Nagasaki University, tend to be biased? An answer warrants careful consideration. 
     
     FMU has its own guidelines about the use and reporting of the FHMS data and its analytical results, requiring any papers to undergo an "internal review" before submission to journals. The internal review is for FMU to ascertain that papers meet conditions such as "contribution to the primary purpose of the FHMS which is to monitor the long-term health of residents" and "accuracy of the results and interpretations." FMU could scrutinize the results and interpretations according to what is considered "accurate." This might be one of the reasons for an arbitrary study design and writing style commonly seen in the FMU studies. 
     In 2016 the Oversight Committee created the Subcommittee for Data Provision for the Purpose of Academic Research (herein, the data subcommittee) in order to establish guidelines on provision the FHMS data to third-party researchers. Waiving informed consent acquisition for the third-party researchers has been considered due to a sheer difficulty of obtaining informed consent from more than half a million Fukushima residents who have participated in the FHMS. 
     In justifying the waiver, the data subcommittee gives consideration to public interest and consent. The subcommittee presumes that facilitation of broad-based research with the FHMS data is critical for maintaining and improving health of residents. Offering residents an opportunity to opt-out is also on the table, "in consideration of their benefit," a euphemism for "so they are not harmed." 
     Regarding the FHMS data, underlying the issue of harm and benefit is FMU's internal rule not to allow research that might ultimately be harmful to residents. This "unscientific" (and unspoken) rule seems to give a priority to curbing psychosocial and economic effects. What exactly is meant by "harm" is usually not spelled out clearly, but what FMU is concerned seems not so much "harm to residents themselves" as "harm to someone other than residents when effects of radiation exposure are shown."
    As a result, any scientific research that might prove effects of radiation exposure could be rejected during an initial application process or an internal review of the finished product. This explains why some of the FHMS papers published so far on medical issues (other than thyroid cancer) barely mention "radiation exposure," attributing any abnormal findings to to evacuation-related stress. 
     Clearly, FMU's internal rule defies the primary purpose of the FHMS to "monitor the long-term health of residents, promote their well-being, and confirm whether long-term, low-dose radiation exposure has health effects." 
     It should be kept in mind that FMU's internal rule is likely to be applied to domestic and international third-party requests to conduct research using the FHMS data, once data provision commences. (Trial period for data request is scheduled to begin in April 2020.)